Metal oxide semiconductors for gas sensing

The usage of the gas sensor has been increasing very rapidly in the industry and in daily life for various potential applications. In the recent years, metal oxide semiconductors (MOS) become the primary choice for designing highly sensitive, stable, and low‐cost real‐life applications‐based gas sensors due to their inherent physical and chemical properties. Researchers have proposed numerous sensing mechanisms to explain the functionality of MOS‐based gas sensors. In this review, we have comprehensively covered different sensing mechanisms used for MOS. We have also discussed different parameters affecting the sensitivity and selectivity of the gas sensors. Moreover, the different techniques used to enhance the gas sensing response of MOS‐based sensors are also extensively covered. And finally, we give our prospective on recent opportunities and challenges on the future applications of MOS‐based gas sensors.


INTRODUCTION
In today's world, we want to expand the limit of our perceptions. To broaden the limit, a large number of sensors are developed. Gas sensor is a device that can detect a specific gas even if the amount or concentration of gas is very small. 1,2 Sensor can help us in identifying the information in our surrounding which is relevant to us and also convert this information into electric signals. 2,3 Gas sensors have been widely used in monitoring industry and domestic environments. 4 These sensors have several advantages such as low power consumption, small size, low cost, and high reliability. The demand of gas sensors is increasing day by day; hence, we need the gas sensors with high sensitivity and better selectivity. 5,6 The sensing material is one of the deciding factors about the performance of gas sensors. The gas sensing industry is dominated by metal oxide semiconductors (MOS) to fulfill the requirement of ideal gas sensors. MOS-based gas sensors have many applications such as environmental monitoring, fire detection, detection of harmful gases in mines, home safety, traffic safety, and healthcare, as shown in Figure 1. 3,7,8 In the recent years, MOS gas sensors are one of the most researched groups of sensors particularly with the nanoscale size of 1-90 nm due to their size dependent properties. The ratio between surface area and their volume increases drastically with decrease in material size. Moreover, the size and materials geometry affected the movement of electrons and holes in the nanomaterials. 5,9 Over the past few years, a wide variety of gas sensors (catalytic-type, electrochemical-type, and MOS type) emerges based on distinct sensing mechanism and sensing materials. Among all the gas sensing materials, MOS sensors performance is higher because of their unique structures, physical, and chemical properties. MOS-based

Sensitivity
Gas sensor sensitivity is defined as the ability to sense gases. The slope of the response curve (also known as the calibration curve) is typically used to determine the sensitivity. The response curve is a plot between the device response and concentration of gas. 24,25 A steeper slope indicates high sensitivity, while a moderate slope signifies a lower sensitivity. In general, the normalized ratio of response signal over baseline is called the sensitivity of the gas sensors.

Selectivity
The selectivity of a gas sensor can be explained as the ability to sense target gas in the presence of other gases. Ideal gas sensor has high selectivity indicating that it mainly senses the target gas and neglecting the other interfering gases. Hence, high selectivity confirms that the sensor gives the accurate information about the existence and concentration of gases.

Stability
Stability indicates the ability of a sensor to produce reliable results over a period of time. MOS-based gas sensors have low stability that leads to the undesired result or false alarms. To some extent stability can be improved by lowering the operating temperature.

Response time
The time taken by the gas sensor to reach 90% of the saturation value after the triggering is called response time.
Logarithmic curve and error function are the common approximations used for the transient response shape. 22

Recovery time
The time taken by the gas sensor to reach 90% of the initial value after "off" the triggering is called the recovery time.

Operating temperature
Ideally, the room temperature-based gas sensors are preferred in most of the practical applications due to their less power consumption, high durability, and easy portability. However, in most of the MOS-based gas sensors, high temperature is required to activate the absorption/desorption process at the sensing surface.

NANOSTRUCTURES FOR GAS SENSING
The morphologies and dimensions of the grown metal oxides strongly influence the sensing performance of gas sensors. Mixed dimensional MOSs are proved to be the primary study materials for sensing different gas analytes. When one or more than one dimension of the nanostructures is equivalent to or less than the Debye length, the gas sensor's response changes dramatically. 26 By developing a new diffusion-reaction equation, Yuan and co-workers examined the effects of grain radius on the transient conditions of gas molecule transmission and the sensitivity of gas molecules. 27 The theoretical investigation demonstrated that with an increase in surface-to-volume ratio more active sites on the surfaces participate in the gas molecule interaction, providing higher performance.

0D Nanostructure
The term "0D nanostructured materials" describes materials that have all of their dimensions of the order of nanoscale range. The 0D nanostructure-based gas sensors include MOSs, noble metals, and quasi-0D quantum dots (QDs), as listed in Table 1. Jaballah and coworkers synthesized the hexagonal structure of a wurtzite aluminum-magnesium co-doped  53 zinc oxide (AZMO) nanostructure using an altered sol-gel technique. 28 Compared to undoped ZnO, AMZO screen printed films demonstrated a significant increase in the sensing performance of the sensor toward CO and CO 2 gas molecules. The improved sensor's responsiveness was ascribed to the perfect union of distinct characteristics such as reduced particle size, crystal quality, and donor defect states. Wang et al. have also studied the sensing ability of ZnO nanoparticles upon exposure to H 2 S molecules. 54 The ZnO nanoparticles were altered with Ag nanoparticles to obtain a high sensitivity of 298 upon exposure to 10 ppm of H 2 S. The improved performance was primarily due to spillover and electron sensitization effects.

1D Nanostructure
When any two dimensions are in the nanoscale range, then it comes into the 1D nanomaterial category. These 1D nanomaterials include nanorods, nanowires, nanotubes, etc., and have appealing topologies and morphologies for gas sensing applications, as summarized in Table 1. Tian et al. have shown high sensitivity and fast response of ZnO Nanorods-based gas sensors toward benzene and ethanol. 55 The distinct surface structure due to more exposed facets was the prime factor in improving the sensing performance. Moreover, doping and surface functionalization of MOSs could further improve the performance of electronic devices. For instance, Lee et al. have demonstrated improvement in the sensing performance of SnO 2 nanorods-based sensors through doping of noble nanoparticles. 56 The doped nanostructure showed a 6 times improvement as compared to the pristine SnO 2 -based sensor upon exposure of 1000 ppm H 2 at 300 • C.

2D Nanostructure
In 2D nanostructures, any one of the dimensions is in the nanoscale range. These materials possess stronger bonds in-plane while they have weak van-der Waals bonding between the monolayers. The 2D nanostructure morphology such as nanosheets, nanodisks, and nanoplates mainly consists of graphene, metal oxide, hexagonal boron nitride (hBN), and metal dichalcogenides nanomaterials, as listed in Table 1. They attract the researchers' attention due to their interesting surface chemistry and quantum confinement effects. Moreover, the versatile physics of 2D materials facilitates them to form heterojunctions with other dimensional materials without considering their lattice constants. Liang et al. have reported atomically thin 2D WO 3 nanosheets for xylene sensing. 57 The ∼4.9 nm thin WO 3 nanosheet showed high sensing performance due to a large number of exposed crystal facets and a high surface-to-volume ratio. Jang et al. have also demonstrated a 2D SnO 2 /CoO x heterogeneous nanosheet-based chemiresistive sensors. 58 The sub-10 nm thick nanosheets showed exceptionally high sensitivity and selectivity upon HCHO exposure due to high porosity and small grain size. Metal-organic frameworks (MOFs) are newly introduced 2D materials that have attracted extensive researchers' interest. MOFs are a combination of metallic compounds and organic ligands. To create 1D, 2D, or 3D nanostructures, metal ions/clusters are linked to organic ligands to produce different morphologies based on MOFs. 59 MOFs are a specific type of coordination polymer with tunable porous structures. MOFs are classified as being either microporous, mesoporous, or macroporous based on the diameter of the pores. Typically, the last two categories are amorphous in nature. MOFs showed a very high specific surface area and excellent thermal stability ranging from 250 • C to 500 • C. MOFs offer a large number of active sites for the gas molecules to get adsorbed as compared to other porous materials due to their very large surface area. Ali et al. have demonstrated a MOFs-based highly sensitive flexible H 2 S gas sensor. 60 The sensor could detect as low as 1 ppm of H 2 S even at room temperature. The improved sensing ability was achieved due to the integration of MOF-5 microparticles into the chitosan organic membrane. The variety of structural frameworks and the versatility of the pores make this category of materials ideal for a wide range of gas sensing applications.
In addition to MOFs, MXenes have also shown their great potential for gas sensing applications. MXenes are hydrophilic in nature and the combination of transition metal carbides, nitrides, and carbonitrides that were found in 2011 at the Drexel University. The general formula of these materials is M n+1 X n T x , where M, X, and T x denote an early transition metal (Ti, V, Nb, Mo, Cr, Ta, or Hf), C or N, surface terminations (typically = -O, -OH, -F, or -Cl), and n varies from 1 to 4. 61 MXenes are very helpful in a wide range of applications including energy storage, optoelectronic, biomedical, and environmental due to their distinctive properties. Hydrofluoric acid is used to selectively etch layers of transition metal carbides and carbonitrides from the MAX phases to create the first generation MXene. Since then, several synthesis techniques have been created, such as selective etching in a solution of fluoride salts and other acids, nonaqueous etchants, and molten salts, enabling the synthesis of novel MXenes with improved control over their surface chemistries.
In the recent years, MXenes emerge to be one of the most effective gas sensing materials due to their high conductivity, surface reactivity, biocompatibility, and stability. In a very recent study, our group extensively discussed the chemical detection capabilities of MXenes. 62 We have also highlighted various ways such as surface functionalization, Schottky barrier modulation, and heterojunctions formation for improving the sensing ability of the MXenes-based gas sensors. MXenes and their nanocomposite gas sensors showed promising results for detecting NO 2 , H 2 , CH 4 , and VOCs.

3D Nanostructure
Nanostructures with low dimensions, such as 0D, 1D, and 2D, have demonstrated considerable and promising results upon exposure to various gas molecules. However, some low-dimensional nanomaterials were unable to maintain their excellent dispersibility and structural stability, leading to degradation in their electrical characteristics (such as sensitivity, response time, and repeatability for gas detection). To overcome this limitation, low-dimensional particles were used to create a 3D nanostructure (all three dimensions are out of the nanoscale range), which can provide superior gas sensing performances, as summarized in Table 1. The geometry of 3D MOSs is typically porous or hierarchical with a less flocculated structure. The 3D MOSs nanomaterials have more precise control over morphologies and pore configurations than other dimensional nanomaterials. Hassan et al. fabricated 3D graphene aerogel hybrid materials decorated with macroporous ZnO nanoparticles (NPs). 63 This macroporous 3D hybrid design, when employed as NO 2 gas sensors for detecting low concentrations of NO 2 gas molecules, showed a very high response. In addition, it also possesses rapid response-recovery time with superior selectivity and repeatability behavior at room temperature. This improved sensing ability was mainly ascribed to enhancement in the conductivity, strong interface between p-graphene and n-ZnO, and a greater number of active sites available at the graphene surface.

GAS SENSING MECHANISM
Based on the operating temperature, MOS can be classified into two categories: the first category involves the materials which follow the surface conductance effects, while the second category comprises the materials which follow the bulk conductance effects. 5 The first category belong to the oxides operated at low-to-moderate temperature (<600 • C), whereas the second category works at very high temperature (>700 • C). 5,64 The operating temperature also defines the mechanism by which these materials functions. Among popular MOS, SnO 2 , and ZnO are some of the oxides which belongs to the first category and called as surface conductance materials. Bulk conductance effect is slow at the lower temperature and change in conductance occurs because of the adsorption and removal of oxygen at the surfaces. 64 Some examples of the bulk conductance materials (BCM) are TiO 2 , Nb 2 O 5 . These BCM belongs to the second category and only respond to variation in the partial oxygen pressure at higher temperature (>700 • C). They also showed equilibrium between the atmosphere and bulk stoichiometry. 64 MOS-based gas sensors detect the presence of gases primarily by recording the changes occurring in the electric properties due to the target gas which is in contact with the sensor. The sensing mechanisms explain the physics affecting the electric properties of a gas sensor upon exposure of target gas molecules. The MOS-based gas sensing mechanisms can be categorized at microscopic and macroscopic levels. The microscopic perspective explains various mechanism such as Fermi-level control theory and charge carrier depletion layer theory. While the macroscopic mechanism focuses on the relationship between the gases and materials. The macroscopic level involves adsorption/desorption processes, bulk resistance control mechanism and gas diffusion control mechanisms. In most of the MOS-based gas sensors, macroscopic level is used to explain the gas sensing mechanisms.
When different gases come in contact with the MOS, the conductivity and work function of the MOS changes significantly. Most of the gas sensing mechanisms work on the adsorption/desorption principle. The adsorption/desorption processes explain the physical or chemical changes in the sensor behavior when the gas is in contact with the sensing surface. Upon exposure of target gas molecules, resistance of the material changes because of the concentration of charge carrier changes in a significant manner. The carrier concentration changes ascribed to chemisorption and physisorption process of target gas molecules and ambient oxygen molecules.
Chemical Adsorption/Desorption is one of the dominant gas sensing mechanisms and strongly affect most of the MOS-based gas sensing devices. When gas comes into direct contact with the sensor, a chemical reaction happens, causing an electrical signal to change. This change might occur due to the presence of targeted gas or ambient oxygen molecules. The oxygen adsorption is one of the most common gas sensing mechanisms and strongly affect most of the MOS-based gas sensing devices. When MOS is brought into air, the oxygen molecules started to adsorb on the material surface and an oxidizing or reducing reaction took place between the atmospheric oxygen and sensing surface. Based on these reactions, some electrical properties or resistance of the sensing material changes considerably. Various kinds of oxygen ions (O − 2 , O − , and O 2− ) are formed in according to the operating temperature after capturing the electrons from the sensing materials. Hence, the conductivity/resistance of the sensing material changes due to change in surface electron density. Upon exposure of reducing targeted gases on n-type MOS, the captured electrons by oxygen species released back to sensing material leading to decrease in resistance, as shown in Figure 3A. 65 The reduction in the value of resistance could further be confirmed by decreased barrier height (Δφ) at the interface. While in the presence of oxidizing gases, the electron density decreases further resulting in increasing the value of resistance. In contrast, upon adsorbing of reducing gases on p-type MOS surface, reduces the hole accumulation layer due to electron-hole recombination process, as depicted in F I G U R E 3 Gas sensing mechanism of (A) n-and (B) p-type semiconducting metal oxide nanostructures (SMONs) upon exposure of reducing gases. Reproduced from Reference 65, with permission from Royal Society of Chemistry Figure 3B. 65 As a result, the surface resistance increases of the p-type MOS structure. However, in the presence of oxidizing gases on p-type MOS, the hole carrier concentration significantly increases due to trapping of electrons by the oxidizing gases and hence the value of resistance decreases.
In addition to chemisorption process, physisorption also play a very important role in understanding of gas sensing mechanism. According to a physical adsorption process, gas molecules is adsorbed at the surface of MOS material via coulomb forces and other different types of intermolecular forces without any chemical changes. Physical adsorption causes a minor difference in the conductivity of MOS materials; hence, this process is not commonly used to explain the gas sensing mechanism. [66][67][68] Humidity sensors are the most common MOS-based sensors, whose sensing mechanism is based on the physical adsorption/desorption processes. 69 Therefore, chemical adsorption is considered as the dominant gas sensing mechanism in the majority of MOS-based gas sensors.
Moreover, the resistance of MOS-based gas sensors could also be changed due to the phase transformations of the gas sensing material. However, this mechanism is only applicable for limited gas sensing materials, such as ABO 3 type MOS composites. 70 α-Fe 2 O 3 and γ-Fe 2 O 3 are the two different types of phase composition of Fe 2 O 3 samples which is prepared by the heating of FeFe(CN 6 ). 71 Based on the two different phases of the material and their combined phases, gas sensing experiments were conducted, and various gas sensing mechanisms of two samples of Fe 2 O 3 were investigated. In the case of α-Fe 2 O 3 , the phase structure was relatively stable; therefore, typical O 2 adsorption model can describe the gas sensing mechanism of α-Fe 2 O 3 . Moreover, due to the changes in the inside structure of γ-Fe 2 O 3 its resistance changes. 72 Hence, α-Fe 2 O 3 and γ-Fe 2 O 3 possess different gas sensing properties despite their identical morphological structures. Gao et al. 73 have also demonstrated that WO 3 hydrogen gas sensing films stability strongly affected by the phase transition. The materials present originally as W 3 O 12 clusters form and by sharing the W-O-W bonds at corner. This whole transition affects the H 2 desorption energy and the HOMO-LUMO gap, resulting in bulk resistance change and improved sensing performance.
In the gas sensing process, materials and gases are the critical components. The chemisorption and physisorption processes are primarily reliant on the chemical and the physical characteristics of that material. In addition, there are some mechanisms largely relying on the process of gas diffusion. Several theories are proposed by scientists indicating that the gas diffusion controls the sensitivity of semiconductor-based gas sensors. [74][75][76] For better understanding of gas sensing properties, Wang et al. have developed a canal and hollow sphere model. 77 When the temperature is less than 150 • C then adsorbed O 2 at the SnO 2 surface was poorly reactive O 2− ions, therefore surface chemical reaction controlled the gas sensing reaction. Now some amount of the target gas is oxidized partially, and the remaining part of target gas diffuses into inner pores and undergo some chemical reactions. When the temperature is in the range of 150-200 • C or greater than 200 • C, then O − and O 2− are the formed, so surface chemical diffusion process is more active now. Hence, complete oxidation and ionization of target gases inside the hollow sphere improves the sensing response. [78][79][80]

METHODS TO IMPROVE GAS SENSING
Gas sensors are mainly used to detect the harmful gases which are threat to human health even at a very low concentration. Therefore, highly sensitive and selective gas sensors are prime need of the electronic industry. [81][82][83][84] The adsorption of gas molecules on the sensing surface could significantly be improved by increasing the number of adsorption sites, increasing the oxygen vacancy, and by enhancing the surface catalytic activity. The sensing material's chemical composition and electronic structure can be modulated by oxygen vacancy engineering. For creating the additional adsorption sites facilitating higher adsorption of target gas molecules, the oxygen vacancies transfer the electrons into the conduction band and cause more electrons to arrive on the surface of the sensor material. Moreover, in the case of electron depletion layer/ hole accumulation layer, 85,86 gas sensing could be improved by barrier reducing and also by the increasing the flow of electrons. Some of the methods to improve the gas sensing are:

Doping with noble metals
Introducing noble metals is one of the most effective and widely used methods for fast catalytic chemical reactions.
Normally, Au, Ag, Pt, and Pd are some of the noble metal dopants used for improving the gas sensing response in the case of MOS sensing materials. There are several steps involved to explain the role of noble metals in improving the sensing response. Firstly, adsorption activation energy is decreased, and adsorption of additional oxygen anions and analyte gas molecules occurs. Secondly, the redistribution of excitons causes the band bending and created a Schottky barrier at the interface, which alters the movement of charge carriers. Lastly, activation energy of the process is lowered by the spillover impact. Due to the addition of the noble metal dopant, the adsorption capacity of gases increases. For example, In the case of W 18 O 49 , when Pd was added then it adsorbed H 2 approximately 900 times of its own volume. Due to this greater adsorption, the effective collision frequency between gas molecules also increased significantly. 87 In other terms, due to this development, the adsorption energy would get decreased after adding the noble metal dopant. Xue et al. demonstrated the sensing behavior of pristine SnO 2 and Pt-SnO 2 upon exposure of CH 4 molecules through first-principle calculations. 88 The adsorption energy of oxygen on the SnO 2 and Pt-SnO 2 are −0.92 and −1.32 eV, respectively, on the exposed crystal planes (110). The obtained results showed that the adsorption of oxygen molecules was more on the noble metal modified SnO 2 as compared to pristine SnO 2 . From Figure 4A,B, it can be clearly visible that adsorption energy is decreased remarkably for all the three adsorption sites when SnO 2 is doped with Pt, confirming that CH 4 is more adsorbed on the surface of Pt-SnO 2 as compared to the SnO 2 . [91][92][93] Moreover, in most of the cases, the work function of MOS gas sensitive materials is less than the noble materials; therefore, redistribution of charge is necessary for achieving the equilibrium. Once the physical contact between metal and MOS is made, the band bending due to charge transfer at the interface formed a Schottky barrier at the junction of MOS and noble metals. The conduction band electrons of MOS sensing material moved into the noble metal particles and form a layer called dipole layer interface ( Figure 4C). This layer at the interface prevents the electron-hole pair recombination process; therefore, the MOS's gas response is increased. This phenomenon is called electronic sensitization. 89,94-96 Figure 4D-F provides an illustration of the factor contributing to the excellent sensing qualities, which has to do with the incorporation of Au-doped atoms. To enhance the chemo-resistive gas sensors' sensing capabilities, metal catalysts such as Au, Pd, Pt, and Ag were frequently utilized. The sensing film functionality was enhanced by catalytic processes on the nanosensors material, which also lead to excellent gas sensor characteristics (such as sensitivity, selectivity, and response and recovery times). [97][98][99][100] The two different methods (i.e., electro-chemical sensitizations) explain why Au nanoparticles have improved sensing performance. 90,101,102 The mismatch in the work functions of Au (i.e., Φ Au = 5.1 eV) and WO 3 (i.e., Φ WO3 = 5.7 eV), which causes the electrical sensitization, leads to a rise in base resistances as seen in Figure 4G. 103 According to Figure 4F, the electron transmission from semiconducting WO 3 to the Au nanoparticles results in an enhancement in resistance, which is implied by an increment in the sizes of the electron depletion zones. Consequently, there were fluctuations in resistance of WO 3 nanosheet due to the treatment of NO 2 analyte gas molecules, which causes enhancement in sensitivity in the WO 3 nanosheets. The second phenomenon, known as chemical sensitization i.e., the catalytic impact, which happens when the Au nanoparticles size is less than 5 nm. Due to the catalytic process and strong contact seen between NO 2 molecules and hydrocarbons occur from the weakening of Au-Au bonds and the modification of d orbitals as Au nanoparticles size reduces. 104 The oxygen molecules were then encouraged to split apart by the Au doped nanocluster in the WO 3 nanosheets, and the resulting reactive oxygen ions. These oxygen ions were diffuse into the WO 3 to act as chemical sensitizers. Due to this active site were increased and this process is known as spill-over effect.
For highly efficient reactions occurring in numerous significant processes (e.g., electrochemistry, gas sensing, and photoreactions), single-atom catalysts (SACs) also provide novel prospects for these reactions. SACs can offer higher reactivity and superior selectivity during the adsorption of gas molecules and electron transmission due to their specific electronic and chemical characteristics. Utilizing SACs increase the sensing selectivity and sensitivity toward particular gases. Recently, it is observed that SACs containing both noble metals (e.g., Pt, Pd, Au, and so on) and non-noble metals (e.g., Mn, Ni, Zn, and so on) can be extremely helpful in enhancing sensing performances. The oxidation state of the metals has been proposed to play a significant role in the electronic as well as chemical properties of the sensing material, which will enhance the gas sensing characteristic. The noble metal catalysts for gas sensors contribute oxidation states that help to enhance the sensing characteristic. The equilibrium of the Fermi levels of the oxidized metal and the semiconductor modifies the carrier concentration of the semiconductor sensing layers. In addition, the noble metal catalyst's spillover effects encourage the separation of molecular oxygen into more active surface atomic oxygen species leading to a very high sensing response. Xu et al. designed Pt conjugated SnO 2 -based resistive gas sensor using the ALD method to detect triethylamine (TEA). 105 A surprising high improvement in the sensor response and a reduction in the operating temperature of the sensor was obtained by single atom Pt catalysts that was coated on the 9 nm SnO 2 layer. Therefore, at an optimum temperature of 200 • C, Pt deposited SnO 2 sensors exhibit the highest response of 136.2 toward 10 ppm of TEA gas molecules. Furthermore, the Pt/SnO 2 -based gas sensor has a quick response time (∼3 s) and recovery time (∼6 s), an ultra-low limit of detection (LOD) of 7 ppb, and excellent sensitivity of 8.76 ppm −1 .

Doping metals other than noble metals
Another effective way to improve the gas sensitivity is addition of heteroatom doping with another metal. When the metal heteroatom is added to the sensing material then it changes the size, porosity and surface area of the MOS, as a result, gas molecules adsorption sites and diffusion paths are modified. 3 In most of the cases, base atom of the MOS is replaced by the metal heteroatom and causes the reduction in grain size. When the size of the grain is less than twice of the length of the Debye then the entire grain size is occupied by the electron depletion layer. Therefore, the MOS gas sensing property is improved. [106][107][108] Now, this is not the case that every heteroatom metal doping increased the gas response. As shown in Figures 5A,B, when atoms of Al, Ga, and Zr are added in the In 2 O 3 , then it improves the response toward formaldehyde. 109 However, when atoms Ti, Mo, and W are added in the In 2 O 3 , then it decreases the material's Fermi level; therefore, this has a negative impact on the gas sensing response. Hence, for using this specific strategy, it is highly recommended that choose the alternative metal heteroatom for doping into the sensing material very carefully for improving the gas sensing performance. 80 A plausible concept was put out to show how a variation in the Fermi level affects the oxide semiconductor's surface sensitivity and this process was well-known sensing process 104,110 and explained by theory related to semiconductor physics. 111,112 This variation in Fermi level is introduced due the absorption of oxygen molecules. Three stages are involved in the sensing process of In 2 O 3 nanomaterial to analyze the formaldehyde gas molecules, as depicted in Figure

Developing heterostructures
Regardless of which sensing material is used, it is nearly impossible to get the maximum of all performance parameters. Therefore, two or more MOS materials are combined to build a heterojunction for overcoming their individual shortcomings. When two distinct materials with different work functions are used for the heterojunction, then the electrons will flow from the lower work function to the higher work function at the interface and this would lead to the electronic depletion and enrichment areas, respectively. 113 The heterojunction gives the appreciative advantage and increase the gas sensing response significantly due to increased catalytic activity and a greater number of available active sites for adsorption of gas molecules. The heterojunctions could be divided into the n-n junctions, p-p junctions, and p-n junctions depending on the type of constituent materials. The charge carrier dynamics at the heterointerface is explained by Figure 6. When p-n heterojunctions are considered, electrons are transferred from n-type to the p-type MOS and holes are moved in the opposite direction till the nanocomposites Fermi level reaches the equilibrium condition. Therefore, resistance will change at the interface of the heterojunction due to the expansion in electron depletion layer. 116 Now, when material is kept in the reducing gas environment, then target gas and adsorbed O 2 react with each other on the materials surface and electrons returned to the n-type MOS conduction band, as illustrated in Figures 6A,B. 114 Moreover, some electrons also enter the conduction band of p-type MOS. Consequently, the recombination of electrons and holes will occur, as a result carrier concentration is reduced on either side of the p-n junction. 65,117,118 Due to the limitation of carrier diffusion, reduction in barrier at the interface is occurred. And apart from that heterojunctions also provides greater adsorption and reaction sites, resulting improved catalytic activity as compared to monomers. 119,120 Gao et al. have demonstrated the H 2 S sensing ability of n-n heterojunction using MoO 3 and SnO 2 MOS. 115 The developed sensor showed high sensitivity and fast response at a lower value of temperature (115 • C) due to larger number of gas adsorption sites at the sensing surface. Upon making physical contact between MoO 3 and SnO 2 , electrons started moving from lower work function SnO 2 to higher work function MoO 3 . The electrons transfer resulted in bend bending having a barrier height φ eff , as shown in Figure 6C. Upon exposure to H 2 S, the trapped electrons by the oxygen ions and targeted gas molecules, returned to the constituent materials leading to a very high value of sensitivity ascribed to reduction in the potential barrier at the interface. The sensing performance of the heterojunction-based sensors could also be optimized through changing the barrier height at the heterointerface. In very recent studies, Lou et al. have discussed several strategies to improve MOSs formaldehyde sensing performance. 121,122 They have shown the effect of sensitization on the sensing performance of SnO2 nanosheets. The hybrid structure showed a response of 4.5 upon exposure to 20 ppm of formaldehyde at 220 • C. The charge movement between SnO 2 nanosheets was explained with the double Schottky barrier model. The high sensitivity was achieved due to a reduction in barrier height upon diffusion of formaldehyde molecules. In another work, Lou et al. have also highlighted the role of SnO 2 /ZnO heterostructure in improving the sensing behavior of the device. 123 The sensing response of the hybrid device became four times that of pristine SnO 2 with 1 ppm formaldehyde at 200 • C. Higher response of the device directly corresponds to the higher concentration of surface absorbed oxygen species in the hybrid device. The remarkable performance of heterojunction-based sensors confirms their suitability particularly to detect toxic gases.
A novel Co 3 O 4 /NiO x heterostructure-based gas sensor synthesized by atomic layer deposition (ALD) also showed promising gas sensing behavior. 124 This novel architecture was designed to enable the controllable management of oxygen vacancies. NiO-ALD cycle optimization controlled the number of oxygen vacancies. Thus, there was a boost in the carrier concentration enhancing the material's conductivity. Due to oxygen vacancy engineering, the gas sensor based on the modified Co 3 O 4 /NiO x sensing layer demonstrates substantially improved performance toward detection of triethylamine gas molecules, at lower working temperature, and possesses excellent sensing characteristics such as higher sensitivity, and lower limit of detection. Similarly, based on the oxygen vacancy engineering and ALD method, Xie et al. manufactured In 2 O 3 /NiO heterostructure-based gas sensor which showed enhanced sensing characteristics toward NO 2 gas molecules. 125 The optimized In 2 O 3 /NiO-based gas sensor shows a higher response (∼532.2) toward a 10 ppm concentration level of NO 2 at 145 • C and a lower limit of detection (∼6.9 ppb).
Bhangare et al. proposed a novel SnO 2 /rGO assembled nanohybrids with a minimum detection limit of 0.5 ppm. 30 The SnO 2 /rGO assembled nanohybrids successfully identified NO 2 gas molecules below their permissible limit. At a working temperature of 200 • C, the 3-fold sensor response was obtained for NO 2 gas molecules having concentrations level of 3 ppm. The nanohybrids showed improved response kinetics with response and recovery timings of ∼10 s and ∼ 40 s, respectively. SnO 2 has a particle size of about 5-7 nm in nanohybrids, which is about twice that of its Debye length (∼3 nm) and boosts both charge transport and sensor response characteristics. In another similar work, SnO 2 -rGO (SR) and N-doped SnO 2 -rGO (SRN) assembled nanohybrids were made using an easy hydrothermal technique by Modak et al. 126 The manufactured assembled nanohybrids' gas sensing capabilities were investigated against different gas molecules. At low operating temperature (∼120 • C) and low gas concentration level (∼0.5 ppm), it was observed that SR and SRN nanohybrids demonstrated improved NO 2 gas sensing response, that is, 55.2% and 84.5%, respectively. Furthermore, SR and SRN demonstrated amazing stability up to 90% over 30 days in addition to outstanding selectivity toward NO 2 . SnO 2 with CuO as a catalyst could also be used for detection of H2S gas molecules. 127

MOS modified with the two-dimensional TMDs materials
The performance of MOS-based gas sensors significantly improves after combining them with TMDs due to their synergistic electronic, chemical, and geometrical effects. The active area of the sensing device for the gas molecules to get adsorbed increases considerably due to the hybridization of MOSs with TMDs. Moreover, the hybridization also facilitates selective diffusion of target gases and efficient carrier transport upon exposure of gas molecules. 61 MX 2 is the general form of transition metal dichalcogenides (TMDs), where M represents the transition metal element (Mo, Ti, etc.) and X refers the chalcogen elements such as S and Se. [128][129][130][131] TMDs is very promising 2D family in the gas sensing applications because of its unique physical and chemical properties. Molybdenum disulphide (MoS 2 ) and MOS composites are one of the most used composites in the gas sensing applications. 86,132,133 Zhang et al. prepared MoS 2 and Co 3 O 4 composite sensor with 1 layer, 3 layers, 5 layers, and 7 layers on a substrate using layer by layer (LBL) self-assembly method. 132 Upon comparing the sensing performance of all those four different layers composite of MoS 2 and Co 3 O 4 , it was observed that the five-layer composite gives the best response upon NH 3 exposure at room temperature. They also synthesized MoS 2 and CuO composites by using the LBL technology. 133 By using this sensor they achieved high sensitivity, quick response and excellent stability for H 2 S gas molecules. 129 The gas sensing mechanism for this composite is attributed to synergistic effect of energy band structure and creation of p-n heterojunction, as shown in Figure 7A,B. For better understanding of gas sensing mechanism between H 2 S and CuO, X-ray diffraction analysis was performed between MoS 2 and CuO nanocomposites. After exposing to NH 3 , some CuS peaks appeared confirming formation of CuS upon CuO and H 2 S reaction. Both CuO and MoS 2 have the different work functions and band gaps as depicted in Figure 7B. Hence, charge transfer will take place until the Fermi level reaches at equilibrium at the heterointerface. The charge distribution forms a depletion layer at the junction and increases the resistance of the composite material. Upon exposure of H 2 S, p-type CuO converted into metallic form and the p-n type heterojunction transformed into metal-semiconductor junction as shown in Figure 7B. This transformation drastically reduces the sensor resistance and hence a very high value of sensitivity was obtained. In addition to MoS 2 , other TMDs such as WS 2 , WSe 2 , SnS 2 , MoSe 2 , and NbS 2 could also be combined with MOSs to improve the sensing ability of the devices. Qin et al. have demonstrated the NH 3 sensing ability of WS 2 at room temperature. 134 They examined the sensing response of bulk to monolayer WS 2 and observed that the sensing response enhanced significantly with a decrease in thickness of WS 2 film. In another study, Hao et al. combined flower-like SnS 2 nanostructures with SnO 2 to increase the number of active sites for NO 2 adsorption. 135 The hybrid structure showed a very high response of 51.1 upon exposure of 1 ppm of NO 2 at 100 • C. The flower-like SnS 2 structure facilitates more NO 2 adsorption and the SnS 2 /SnO 2 heterojunction played important role in enhancing the sensing response of the device. In a very recent study, Zhou et al. fabricated MoO 3 /MoSe 2 hybrid for trimethylamine sensing. 136 The nanocomposite showed high sensitivity, low value of detection limit, fast response, good stability, and exceptional selectivity even at room temperature. The unique flower shaped MoSe 2 morphology and the formation of heterojunction were the main driving factors for the high gas sensing performance.

Light-induced room temperature gas sensing
Light activation is used as a promising approach in sensing applications for oxide layers such as ZnO, TiO 2 , WO 3 , etc [137][138][139] In MOS-based gas sensing processes, high temperature is required for adsorption and desorption of targeted gas molecules. The energy provided by the external heating element could also be supplied by UV light source resulting in less power consumption and room temperature operation of the developed sensor. 140,141 In addition, UV light also decrease the measurement time and used for cleaning the gas sensing surface after exposing with targeted gases. 137 The 1/f noise factor in resistive gas sensors is quite strong and dominates the background noise up to 10 kHz. When potential barriers between the grain's boundaries vary, it leads to the generation of low-frequency noise. The gas sensor's temperature and surrounding atmosphere can be specifically affected by the barrier's variations because it relies on the adsorption and desorption mechanisms. The 1/f noise is shown to rise with decreasing grain size in experimental data. Furthermore, when gas-sensor nanomaterial grain size reduces, the sensitivity of sensor get enhanced and noise fluctuations becomes even more sensitive than that of variations in DC resistance. 142 Smulko et al. have discussed the application of photo activation for improving the gas sensing ability of the sensors at lower value of operating temperature. 143 The change in DC resistance (Rs) of sensor is compared after the increment in heating voltage upon UV irradiation (365 nm) by the help of T5F UV diode. 144 This diode was placed approximately 5 mm away from the sensing layer and its DC current (Id) was controllable up to a maximum of 20 mA. This recorded curve shape depends on the surrounding atmosphere as this was examined for synthetic air and ethanol. It was observed that the recorded data is quite similar for both the cases whether it was UV light or temperature modulation. 143 But one thing is very noticeable that power consumption is lesser in UV light than the power required for temperature modulation. Hence, by using the UV light, sensor can work at lower temperature, which is very appealing characteristic. 143  Figure 8A. 145 The device showed bifunctional selectivity toward triethylamine and NO 2 under different conditions. Moreover, in presence of UV illumination, the sensor displayed a very low limit of detection (8 ppb) for NO 2 at room temperature. The UV illumination assists in the generation of a greater number of electron-hole pairs. Therefore, a high concentration of electrons would be available for NO 2 , as shown in Figure 8B leading to a very high value of sensitivity. In a similar study, Lu et al. have also improved the sensing ability of monolayer WSe 2 through UV activation. 146 The electrons get trapped from WSe 2 upon exposure to NO 2 resulting in an increasing the hole concentration. After UV illumination, the new electron-hole pairs are generated facilitating more NO 2 molecules to get adsorbed.

5.6
Nanostructures with exposed facets Nanostructured exposed crystal facets exhibit large difference in their gas sensing abilities because of their distinct surface properties. The crystal facets with more defect sites possess a higher value of surface energy, therefore they have more active physicochemical features. [147][148][149] Exposed crystal facets of MOS materials have been used very often in the photocatalysis and recently MOS materials have also showed enormous potential in the realm of gas sensing. 150,151 Overall, the improvement in gas adsorption ascribed to greater density value of dangling bonds and coordination of unsaturated oxygen. For example, upon observing every crystal planes atomic structure diagram of ZnO and calculating the densities of dangling bond of every plane, it revealed that the crystal plane (0001) has higher density. 18 Dangling bonds due to the Zn 2+ have unsaturated O 2 coordination on the surface and when the gas sensing reaction starts then target gases and O 2 anions are adsorbed at these locations. Kaneti et al. 152 used density function theory to examine the ZnO sensing response upon exposure to ethanol. They observed that the exposed crystal plane (0001) interacted with the surface oxygen; the H-O bonds are then shortened and reduces the adsorption energy. Hence, in the case of (0001) crystal plane, an excellent gas sensitivity could be obtained. 18 When target gas and sensing materials surface comes in contact, then there is a higher charge transfer and stronger electron interaction between the target gas and the high energy crystal facet resulting in quicker and greater gas response, as shown in Figure 9A,B. 153 To investigate the selectivity of particular gases (such as acetone, ethanol, H 2 , H 2 S, NH 3 , and NO 2 ), the responses of several nanocrystals to distinct gases each having concentration level of 100 ppm at 320 • C were examined as shown in Figure 9C. It was discovered that NS-010 displays a noticeably stronger acetone response when contrasted to other gases, demonstrating superior acetone sensing selectivity, whereas NS-101 (NS denoted TiO 2 nanocrystals with anatase have been fabricated with different facets) displays a substantially lower acetone selectivity. The outcomes showed that the crystal facet tuning can be used to achieve selective recognition for the acetone. At concentration level of 100 ppm acetone response was measured on the TiO 2 nanosheet as sensing materials, and this experiment were conducted continuously for 15 days in order to examine stability as shown in Figure 9D. The obtained results showed a great consistency of the TiO 2 nanostructures, as evident from the fact that both NS-010 and NS-101 sustain about 95% of the initial reaction even afterwards being tested for 15 days.

CONCLUSION AND PERSPECTIVES
The MOS has been the focal point for developing efficient gas sensors due to their inherent characteristics. This review article provides the in-depth knowledge of different types of gas sensing mechanism and different types of methods responsible for improvement of gas sensing. We have also discussed about different kinds of MOS nanostructures such as nanorods, nanobelts, nanofibers, nanosheets, and their composites. The recent progress in sensing mechanism for detecting different kinds of hazardous and flammable gases including NO 2 , NH 3 , H 2 S, H 2 , and volatile organic compounds have also been summarized. The study of MOS-based gas sensor still faces many opportunities and challenges. There are still some questions which cannot be explained by using the existing gas sensing mechanism. Therefore, it is necessary to do more research in this topic for the better understanding of the topic. In this area, more experiments are required using the modern technique such as situ analysis to know about the effect of different type of gas sensing mechanism on the performance of gas sensors. By studying the unexplored physics, we can select the appropriate and accurate gas sensing mechanism in the future.
The chemisorbed oxygen species on the MOS surface slow down the desorption process and requires some external energy to improve the speed of the sensor. Thermal and photo energy appears to be the most widely used external sources to expedite the desorption mechanism. However, these techniques increase the power consumption and restrict the portability of the sensor. Research on the topic of MOS-based gas sensor is increasing day by day, but attention paid to this topic is still not sufficient; therefore, most articles have failed to give accurate information regarding the role of heterojunction in the gas sensing, gas sensing mechanism and the best material for sensing particular gases. AUTHOR CONTRIBUTIONS Neeraj Goel: Conceptualization (equal); formal analysis (equal); supervision (equal); writing -original draft (equal); writing -review and editing (lead). Kishor Kunal: Formal analysis (equal); writing -original draft (supporting); writing -review and editing (supporting). Aditya Kushwaha: Writing -original draft (equal); writing -review and editing (supporting). Mahesh Kumar: Supervision (equal); writing -review and editing (supporting).

CONFLICT OF INTEREST
The authors have no conflict of interest to report.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.